Civic
A battle over open information and algorithmic power is reshaping how millions of Brazilians access political reality.
NewsOnScale Staff
September 17, 2026
Brazil has spent the better part of a decade testing the outer limits of what digital infrastructure can do to a democracy — and to it. From the WhatsApp-fueled misinformation campaigns that defined the 2018 and 2022 election cycles to the January 8, 2023 Capitol-style assault on Brasília, the country has functioned as a live stress test for the relationship between information technology and political stability. Now, with AI systems beginning to mediate how citizens find, interpret, and share civic knowledge, that stress test is entering a new phase.
The core tension is straightforward, even if the technical details are not. Open knowledge ecosystems — public-domain databases, open-source models, government transparency portals, community-edited encyclopedias — represent one pathway for AI development. Closed, proprietary systems controlled by a small number of platforms represent another. Brazil is currently navigating both at once, and the choices being made right now will determine which model wins at scale.
## The Infrastructure Problem Nobody Wants to Talk About
When we talk about AI and democracy, the conversation usually gravitates toward deepfakes, chatbot-generated propaganda, or algorithmic amplification of extreme content. Those are real concerns. But there is a quieter, more structural issue underneath them: who controls the training data, the retrieval systems, and the knowledge bases that AI tools draw upon when a citizen asks a political question.
In Brazil, that question has sharp teeth. Portuguese-language civic data is underrepresented in most major AI training sets. Government transparency portals exist but are frequently inaccessible in machine-readable formats. Civil society organizations that have spent years building open civic databases operate on precarious funding. Meanwhile, large platform companies have the resources to build proprietary knowledge layers — and every incentive to do so.
The result is a two-tier system taking shape in slow motion. Institutions, journalists, and well-resourced campaigns get AI tools trained on rich, curated data. Ordinary citizens get whatever the consumer-facing product serves up, with limited visibility into its sources or its gaps.
## The Open Knowledge Argument
Advocates for open civic infrastructure make a case that deserves serious engagement rather than reflexive agreement or dismissal. Their argument is that democratic legitimacy depends not just on elections but on a reasonably shared information environment — and that AI systems built on proprietary knowledge bases actively erode that shared environment by making it opaque.
This is not a conspiratorial claim. It is an architectural one. If the system that answers your question about a candidate's voting record was trained on data you cannot inspect, updated on a schedule you don't know, and governed by a terms-of-service agreement written by a private company, then you have outsourced a meaningful slice of your civic cognition to an unaccountable actor. Multiply that by millions of users and the aggregate effect on democratic deliberation is not trivial.
Brazil's open-knowledge advocates are pushing for government investment in Portuguese-language civic AI tools built on auditable data, mandatory transparency requirements for AI systems used in political contexts, and structured public access to government datasets in formats that civil society developers can actually use.
## Why This Matters Beyond Brazil
Brazil is large enough — 215 million people, the world's fourth-largest democracy by electorate — that what happens there matters globally. It is also a leading indicator. The disinformation patterns that emerged on Brazilian WhatsApp networks in 2018 showed up in other countries within electoral cycles. The platform moderation fights that consumed Brazilian regulators in 2023 and 2024 are now being replicated across the Global South.
If Brazil develops a workable model for open civic AI infrastructure — or fails to — other democracies will be watching and adapting accordingly.
The uncomfortable reality is that the window for building that open infrastructure is not permanently open. As proprietary AI systems embed themselves deeper into how people navigate political information, the switching costs rise and the open alternatives become harder to bootstrap. Brazil's current moment is not a philosophical debate. It is a procurement decision, a regulatory choice, and a funding allocation — all happening simultaneously, mostly out of public view.
That is exactly the kind of story that tends to get covered only after the window closes.